A multi-pulse seismic exploration structure and a mine ground stress prediction method and device

Abstract

The application relates to the technical field of ground stress prediction, and discloses a multi-pulse seismic detection structure and a mine ground stress prediction method and device, wherein the detection structure comprises a field host, a multi-pulse seismic source, a cable, a three-component detector, a preamplifier, a collection station, an anchor rod tray and an anchor rod; the multi-pulse seismic source is connected with the field host through the cable, and a blast hole is blocked by yellow mud; the three-component detector is rigidly connected with the tail end of the anchor rod through the anchor rod tray, is sequentially connected with the preamplifier, and is connected with the collection station which is connected with the host; a stress meter is arranged in a free section of the anchor rod; the stress meter comprises a ceramic substrate, a temperature compensation sheet, a resistance strain gauge and waterproof glue; the strain gauge and the compensation sheet are fixed to the ceramic substrate through the waterproof glue, and are connected with the collection station through a signal adjustment module and the cable. The application can comprehensively and accurately predict the ground stress in the whole space, and provides direct geological information for geological research and engineering application.

Description

Technical Field

[0001] This application relates to the field of geostress prediction technology, and in particular to a multi-pulse seismic detection structure and a method and apparatus for predicting geostress in mines. Background Technology

[0002] In underground engineering construction, safety is the primary prerequisite, and efficiency is the core objective. However, tunnel construction is poorly adaptable to complex geological conditions (such as fractured zones and collapse columns), and is highly susceptible to accidents such as sudden water inrush, collapses, and large deformations of the surrounding rock. Especially under high ground stress, the risk of disasters increases dramatically, potentially leading to a chain of accidents such as well flooding, pipeline relocation, and obstruction of resource development.

[0003] In-situ stress is the pressure exerted on surrounding rock under natural conditions or the influence of mining activities. It is formed by the combined effects of various factors, including geological structure, the weight of the rock itself, and mining disturbance. It is a key factor affecting mine stability and safety. As the depth of mineral resource mining in my country increases and geological conditions become more complex, safety issues caused by high pressure in deep formations are becoming increasingly prominent.

[0004] Currently, existing technologies mainly utilize single-pulse seismic detection technology for full-space detection in mines. This technology mainly relies on the nonlinear relationship between seismic wave propagation velocity and the mechanical parameters and stress state of the underground medium. By analyzing the changes in the transmission, reflection, and refraction characteristics of seismic waves during propagation, the geological conditions of the entire mine space can be indirectly determined. With the increasing requirements for explosive control and the demand for green mining, the detection technology using hammer impact as a seismic source has gradually gained attention. However, the hammer impact method has obvious limitations: (1) Insufficient detection depth: Due to the small single excitation energy (usually <500J), the effective detection distance in complex geological environments is usually less than 80m; (2) Weak noise resistance: In broadband noise environments, the effective signal of single-pulse detection is covered by more than 60%. Information needs to be collected repeatedly (>3 times) for superposition processing, which greatly reduces efficiency; (3) Only structural surface information can be obtained. The inverted velocity magnitude is used to qualitatively analyze the ground stress, but the quantitative ground stress intensity cannot be obtained.

[0005] In summary, single-pulse seismic detection methods have significant limitations in directly acquiring geological information. These shortcomings greatly restrict the effectiveness of this technology in complex geological environments, especially in deep mines and complex strata. To address these issues, new detection methods and technologies are urgently needed. Summary of the Invention

[0006] The purpose of this application is to provide a multi-pulse seismic detection structure. By employing multi-pulse coherence enhancement and intelligent noise suppression to optimize signal processing, the problems of low signal-to-noise ratio, poor resolution, and weak anti-interference capability are solved, thereby obtaining more accurate imaging structural surface results and inverted wave velocities. Based on this, utilizing the characteristic that the magnitude of the in-situ stress at the anchor bolt is the standard solution, the initial data (including imaging structural surface results and wave velocity data) and boundary constraints (stress at the anchor bolt) are input into a deep learning model to perform in-situ stress inversion calculations, deriving the in-situ stress distribution throughout the underground space, thus achieving a more accurate prediction of in-situ stress across the entire space.

[0007] To achieve the above objectives, the following technical solution is adopted:

[0008] In a first aspect, this application provides a multipulse seismic detection structure, including a field host, a multipulse source, cables, a three-component geophone, a preamplifier, an acquisition station, an anchor tray, and anchor bolts; wherein:

[0009] The multi-pulse seismic source is connected to the field host via a cable, and its blast holes are sealed with yellow mud sealing components.

[0010] The three-component detector is rigidly connected to the end of the anchor rod via the anchor rod tray, and is sequentially connected to the preamplifier and the acquisition station, and finally connected to the field host.

[0011] The free section of the anchor rod is equipped with a stress gauge, which includes a ceramic substrate, a temperature compensation plate, a resistance strain gauge, and waterproof adhesive.

[0012] The resistance strain gauge and the temperature compensation plate are fixed to the ceramic substrate by the waterproof adhesive and connected to the signal adjustment module by a cable;

[0013] The signal adjustment module is connected to the acquisition station via a cable.

[0014] Furthermore, the distance between the blast hole of the multi-pulse seismic source and the loosened zone of the surrounding rock is greater than 0.5m, the drilling direction is perpendicular to the tunnel wall, the hole depth is controlled between 2.0 and 2.5m, the angle deviation does not exceed ±5°, the hole position is 1.2 to 1.5m above the floor, avoiding the support anchor bolts with a clearance distance of not less than 0.5m, the charge of the electronically coded blasting source is 50±2g, and the hole is sealed with compacted yellow mud with a sealing density of not less than 1.8g / cm³. 3 The seismic source is deployed on one side, with a standard spacing of 20-30m, and the spacing is increased to 15m in fault and fracture zone areas.

[0015] Furthermore, the three-component geophone is connected to the anchor plate via a rigid transition band, and its orientation is fixed as follows: the x-axis points in the excavation direction, the y-axis is horizontal and perpendicular to the rock wall, and the z-axis is vertically downward.

[0016] Furthermore, the acquisition station achieves synchronous triggering with the multipulse source, three-component detector, and signal adjustment module through fiber optic / network protocol dual-mode time synchronization, with a delay of ≤0.1μs, and synchronously stores seismic waveform data and anchor strain data.

[0017] Furthermore, the ends of the anchor bolts are ground smooth and penetrate ≥3m into hard rock layers, accounting for ≥60%.

[0018] Secondly, this application provides a multi-pulse seismic mine stress prediction method based on the multi-pulse seismic detection structure described above, the method comprising:

[0019] Detection Preparation: Based on the spatial structure and geological conditions of the mine roadway, determine the layout and excitation parameters of the multi-pulse seismic source; the excitation parameters include excitation intensity and multi-pulse data; distribute the anchor bolts along the shoulders and bottom of the roadway on both sides, place the stress gauge at the front end of the anchor bolt, and place the high-sensitivity geophone at the end of the anchor bolt and expose it, ensuring close contact between the anchor bolt and the surrounding rock, and between the geophone and the anchor bolt; connect the three-component geophone and the stress gauge end to the preamplifier and signal adjustment module respectively, and perform system connection and debugging; record the spatial geometry information of the mine roadway, the type of surrounding rock, structural features, anchor bolt layout and depth, and store them in the geological database;

[0020] Signal acquisition: A multi-pulse seismic source channel wave signal is generated according to the determined excitation parameters. A three-component geophone receives and records the seismic wave signal. Multiple excitations are performed in each acquisition cycle and complete seismic waveform data is stored. The strain signal of the anchor bolt is recorded in real time by a stress gauge to obtain strain information. The on-site host synchronously processes and stores the strain data. The complete seismic waveform data and strain information are synchronously recorded through network protocol time synchronization.

[0021] Data processing: Based on the complete seismic waveform data, the internal structural surfaces and adverse geological bodies of the tunnel are located using reverse time migration, and the velocity distribution map of the underground structural surfaces is constructed using tomographic inversion algorithm; the strain information is preprocessed by median filtering, and the local stress value of the rock mass around the anchor is inferred from the stress-strain relationship based on the elastic parameters and geometric characteristics of the anchor, and the measured stress data of multiple anchor measuring points are summarized.

[0022] In-situ stress prediction: Using the inverted seismic wave velocity distribution, imaging structural surface, and anchor stress data as input features, a semi-supervised learning deep neural network model is used to construct the mapping relationship between anchor single-point stress, medium wave velocity, and imaging structural surface. With anchor single-point stress and corresponding wave velocity as constraints, and incorporating the imaging direction of the structural surface, a quantitative prediction of in-situ stress in the entire underground space is established.

[0023] Furthermore, based on the complete seismic waveform data, reverse time migration is used to locate the internal structural surfaces and adverse geological bodies of the tunnel. A velocity distribution map of the underground structural surfaces is constructed using a tomographic inversion algorithm, including:

[0024] The seismic wave signal is subjected to noise filtering, and the anisotropy parameters of the dispersion curve are extracted;

[0025] The reflected wave components are extracted based on the anisotropic parameters of the reflected wave dispersion curve, and an initial wave velocity model is constructed using the first arrival travel time tomography method.

[0026] Based on the initial velocity field, the wave velocity distribution is corrected using formula (1), and fine imaging of the structural surface is achieved using reverse time migration technology:

[0027]

[0028] In the formula, s obs(t) : Full waveform, including direct wave, reflected wave and scattered wave;

[0029] forward operator;

[0030] β: Regularization weight;

[0031] T: Total duration of seismic wave acquisition;

[0032] v0: Initial wave velocity model;

[0033] v: v = v0 + Δv, where v is the total velocity field to be inverted, and Δv is the velocity increment;

[0034] v(x): Final wave velocity field.

[0035] Furthermore, it also includes a controller that, based on the anchor bolt's elastic parameters and geometric characteristics, uses the stress-strain relationship to infer the local stress value of the rock mass surrounding the anchor bolt. The formula is as follows:

[0036] σ=E*ε (2)

[0037] In the formula, σ is stress, E is elastic modulus, and ε is strain.

[0038] Furthermore, using the inverted seismic wave velocity distribution, imaging structural surface, and anchor stress data as input features, a semi-supervised deep neural network model is employed to construct the mapping relationship between anchor stress at a single point, medium wave velocity, and the imaging structural surface. With anchor stress at a single point and corresponding wave velocity as constraints, and incorporating the imaging direction of the structural surface, a quantitative prediction of subsurface stress across the entire space is established, including:

[0039] Constructing a semi-supervised learning deep neural network model;

[0040] The inverted seismic wave velocity distribution, imaging structure surface, and anchor stress data are used as input features. The magnitude of the ground stress at the monitored anchor and the imaging structure surface are used as the supervision basis, while the ground stress at the non-monitored anchor is learned based on the imaging structure surface and wave velocity, thus realizing semi-supervised ground stress prediction.

[0041] A three-dimensional spatial interpolation method is used to make the predicted point stress data continuous in space and construct the geostress distribution field.

[0042] Furthermore, the semi-supervised learning deep neural network model is a convolutional neural network or a graph neural network.

[0043] Thirdly, this application provides a multi-pulse seismic mine stress prediction device, comprising the multi-pulse seismic detection structure, signal acquisition module, data processing module, and stress prediction module as described above, wherein:

[0044] The signal acquisition module is configured to generate multi-pulse seismic source slot wave signals according to determined excitation parameters. A three-component geophone receives and records the seismic wave signals. In each acquisition cycle, multiple rounds of excitation are performed and complete seismic waveform data is stored. Anchor bolt strain signals are recorded in real time by a stress gauge to obtain strain information. The on-site host synchronously processes and stores the strain data. The complete seismic waveform data and strain information are synchronously recorded through network protocol time synchronization.

[0045] The data processing module is configured to locate the internal structural surfaces and adverse geological bodies of the tunnel using reverse time migration based on the complete seismic waveform data, construct the velocity distribution map of the underground structural surfaces using tomographic inversion algorithm, perform median filtering preprocessing on the strain information, and infer the local stress value of the rock mass around the anchor bolt based on the elastic parameters and geometric characteristics of the anchor bolt through the stress-strain relationship, and summarize the measured stress data of multiple anchor bolt measuring points.

[0046] The geostress prediction module is configured to take the inverted seismic wave velocity distribution, imaging structural surface, and anchor stress data as input features, and use a semi-supervised learning deep neural network model to construct the mapping relationship between anchor single-point geostress, medium wave velocity, and imaging structural surface. With anchor single-point geostress and corresponding wave velocity as constraints, and incorporating the imaging direction of the structural surface, a quantitative prediction of geostress in the entire underground space is established.

[0047] The beneficial effects of this application are:

[0048] 1) This application utilizes a multi-pulse seismic detection method combined with anchor bolt information, achieving superior performance compared to traditional single-pulse techniques in terms of detection range, signal-to-noise ratio, and imaging structural surface accuracy. Furthermore, by using highly accurate imaging structural surface results and inverted wave velocities as initial conditions, and anchor bolt stress information as boundary constraints, quantitative geostress estimation is achieved. Therefore, this application can comprehensively and accurately predict geostress across the entire space, providing direct geological information for geological research and engineering applications.

[0049] 2) This application, based on traditional single-pulse acquisition, adopts a multi-pulse continuous generation seismic wave detection method. Coherent processing technology is used to superimpose multiple pulse signals, effectively improving the signal-to-noise ratio and detection range. During signal superposition, noise is randomly distributed and cancels each other out, while the real seismic signal is correlated, thus amplifying and enhancing the real seismic signal. This process effectively improves the signal-to-noise ratio and detection range of existing hammer-source seismic data. Based on this, the application incorporates the ground stress at the anchor bolt as a boundary constraint, establishing an unsupervised learning ground stress prediction theory with medium wave velocity and imaging structural surface results as the link. This maps the limited ground stress information to the entire space, overcoming the shortcomings of traditional techniques in qualitatively estimating ground stress and achieving quantitative prediction of ground stress. Attached Figure Description

[0050] Figure 1 A schematic diagram of a multi-pulse seismic detection structure according to an embodiment of this application is shown.

[0051] Figure 2 A structural diagram of the anchor bolt in a multi-pulse seismic detection structure according to an embodiment of this application is shown.

[0052] Figure 3 A flowchart illustrating an overall process for predicting ground stress in mines using multi-pulse seismic methods according to an embodiment of this application is shown.

[0053] Figure 4 A simplified flowchart of a multi-pulse seismic mine stress prediction method according to an embodiment of this application is shown.

[0054] Figure 5 A flowchart illustrating the detection preparation process in a multi-pulse seismic mine stress prediction method according to an embodiment of this application is shown.

[0055] Figure 6 A flowchart illustrating the signal acquisition process in a multi-pulse seismic mine stress prediction method according to an embodiment of this application is shown.

[0056] Figure 7 A data processing flowchart is shown in a multi-pulse seismic mine stress prediction method according to an embodiment of this application.

[0057] Figure 8A schematic diagram of the geostress prediction results according to an embodiment of this application is shown.

[0058] Figure 9 A structural diagram of a multi-pulse seismic mine stress prediction device according to an embodiment of this application is shown.

[0059] Figure label:

[0060] 1. Field host; 2. Multi-pulse source; 3. Cable; 4. Three-component detector; 5. Preamplifier; 6. Acquisition station; 7. Anchor bolt tray; 8. Anchor bolt; 9. Yellow mud sealing component; 10. Ceramic substrate; 11. Temperature compensation plate; 12. Resistance strain gauge; 13. Waterproof adhesive; 14. Signal adjustment module. Detailed Implementation

[0061] The following specific examples illustrate the implementation of this application. Those skilled in the art can easily understand other advantages and effects of this application from the content disclosed in this specification. This application can also be implemented or applied through other different specific embodiments, and various details in this specification can also be modified or changed based on different viewpoints and applications without departing from the spirit of this application. It should be noted that, unless otherwise specified, the following embodiments and features in the embodiments can be combined with each other.

[0062] The specific implementation methods of this application will be further described in detail below with reference to the accompanying drawings and embodiments.

[0063] Example 1:

[0064] This application provides a multi-pulse seismic detection structure, such as... Figure 1 and 2 As shown, the multipulse seismic detection structure includes a field host 1, a multipulse source 2, a cable 3, a three-component geophone 4, a preamplifier 5, an acquisition station 6, an anchor tray 7, and an anchor 8. The multipulse source 2 is connected to the field host 1 via the cable 3, and its borehole is sealed with a mud sealant 9. The three-component geophone 4 is rigidly connected to the end of the anchor 8 via the anchor tray 7, and sequentially connected to the preamplifier 5 and the acquisition station 6, ultimately connecting to the field host 1. A stress gauge is installed on the free section of the anchor 8. The stress gauge includes a ceramic substrate 10, a temperature compensation plate 11, a resistance strain gauge 12, and waterproof adhesive. The resistance strain gauge 12 and the temperature compensation plate 11 are fixed to the ceramic substrate 10 with waterproof adhesive 13 and connected to a signal adjustment module 14 via the cable 3. The signal adjustment module 14 is connected to the acquisition station 6 via the cable 3.

[0065] In this embodiment, the field host 1 serves as the core control center, coordinating the excitation timing of the multi-pulse seismic source 2 and controlling the data storage process of the acquisition station 6. It can store computer programs to execute the multi-pulse seismic mine stress prediction method described in subsequent embodiment 2. The multi-pulse seismic source 2 acts as a seismic wave excitation source, generating multiple rounds of continuous pulse seismic waves (trough waves). The charge amount is controlled electronically, and energy is efficiently coupled to the rock mass via the mud sealing component 9. The cable 3 serves as a signal and power transmission channel, connecting the seismic source 2, the geophone 4, the strain gauge 12, and the host 1, transmitting control commands, synchronization signals, and raw data. The three-component geophone 4 is used for three-dimensional wavefield acquisition. It is rigidly fixed to the end of the anchor bolt 8, directly contacting the surrounding rock, and capturing seismic reflected waves in the X / Y / Z directions in real time. The preamplifier 5 is used for signal pre-enhancement, reducing noise and amplifying the weak reflected wave signal output by the geophone 4 to improve the signal-to-noise ratio. Acquisition station 6 is used for data synchronization and aggregation. It receives and temporarily stores the post-seismic wave signals and the raw data from strain gauge 12. Multi-device synchronization (delay ≤ 0.1μs) is ensured through fiber optic / NTP dual-mode time synchronization. Anchor bolt tray 7 serves as a mechanical transition interface, providing a rigid connection structure to securely mount the geophone 4 to the end of anchor bolt 8, eliminating signal attenuation caused by loosening. Anchor bolt 8 serves as a dual-function carrier, embedding itself into the rock mass to provide tunnel support and acting as a sensor base. Its end rigidly fixes the geophone 4, while the free section is fitted with strain gauge 12. Yellow mud sealing component 9 serves as an energy coupling medium, used to compact and fill the source borehole, ensuring efficient transmission of seismic waves to the rock mass. Ceramic substrate 10 serves as the strain sensing substrate, providing a flat, insulated mounting surface for strain gauge 12, adapting to the high-temperature and humid underground environment. Temperature compensation element 11 is used to suppress environmental interference, eliminating the drift effect of temperature changes on strain gauge 12 data and improving stress measurement accuracy. The resistance strain gauge 12 is used to achieve micro-strain sensing. It is symmetrically attached to the free section of the anchor bolt 8 to monitor the anchor bolt deformation caused by the surrounding rock stress in real time. The waterproof adhesive 13 is used for sealing and fixing. By encapsulating the strain gauge 12 and the temperature compensation plate 11 on the ceramic substrate 10, it prevents water vapor corrosion from causing short circuits or failures. The signal adjustment module 14 is used to condition the strain signal. By filtering (such as median filtering) and amplifying the raw data from the strain gauge 12, it outputs a standardized stress signal to the acquisition station 6.

[0066] In practice, the on-site host 1 controls the multi-pulse source 2 via cable 3 to generate multiple rounds of continuous seismic waves according to preset parameters. The energy is efficiently coupled to the rock mass via a mud-sealing component 9. The seismic waves propagate in the surrounding rock, generating reflected waves upon encountering structural surfaces (such as faults or fracture zones), while the direct waves carry shallow geological information. A three-component geophone 4 is rigidly connected to the end of the anchor bolt 8 via an anchor bolt tray 7, directly contacting the rock mass and capturing reflected wave signals in the X, Y, and Z directions in real time. These signals are then amplified and noise-reduced by a preamplifier 5 before being transmitted to the acquisition station 6. Changes in surrounding rock stress cause micro-strain in the anchor bolt 8. A resistance strain gauge 12 and a temperature compensation gauge 11 are fixed to a ceramic substrate 10 with waterproof adhesive 13, eliminating temperature drift. The acquired strain signals are then filtered and conditioned by a signal adjustment module 14 before being synchronously input to the acquisition station 6.

[0067] In some embodiments, the distance between the blast hole of the multi-pulse seismic source and the loosened zone of the surrounding rock is greater than 0.5m, the drilling direction is perpendicular to the tunnel wall, the hole depth is controlled between 2.0 and 2.5m, the angle deviation does not exceed ±5°, the hole position is 1.2 to 1.5m above the floor, avoiding the support anchor bolts with a clearance distance of not less than 0.5m, the charge of the electronically coded blasting source is 50±2g, and the hole is sealed with compacted yellow mud with a sealing density of not less than 1.8g / cm³. 3 The seismic source is deployed on one side, with a standard spacing of 20-30m, and the spacing is increased to 15m in fault and fracture zone areas.

[0068] It should be noted that the borehole parameters mentioned in this embodiment (including but not limited to: distance from the loosened zone 0.5m, hole depth 2.0–2.5m, charge amount 50±2g, and yellow mud sealing density 1.8g / cm³) 3 The deployment spacing (e.g., 20–30m / 15m) is merely an exemplary implementation scheme used to illustrate the technical principles. In practical applications, adjustments can be made based on mine geological conditions, tunnel dimensions, and detection targets. These specific values ​​do not constitute a limitation on the scope of protection of this application.

[0069] In some embodiments, the three-component geophone 4 is connected to the anchor plate 7 via a rigid transition band, with its orientation fixed as follows: the x-axis points in the excavation direction, the y-axis is horizontal and perpendicular to the rock wall, and the z-axis is vertically downward. The connection between the geophone 4 and the anchor plate 7 via the rigid transition band eliminates interference from flexible vibrations and ensures that the seismic wave signal is transmitted to the sensor without loss.

[0070] In some embodiments, the acquisition station 6 achieves synchronous triggering with the multipulse source 2, the three-component detector 4, and the signal adjustment module 14 through fiber optic / network protocol dual-mode time synchronization, with a delay of ≤0.1μs, and synchronously stores seismic waveform data and anchor strain data.

[0071] In some embodiments, the end of the anchor bolt 8 is ground flat and penetrates into hard rock layers for ≥3m, accounting for ≥60%.

[0072] Example 2:

[0073] This application provides a multi-pulse seismic method for predicting in-situ stress in mines. This method can be based on a multi-pulse seismic detection structure as described in Embodiment 1, which can provide the necessary data for implementing the prediction method. Figure 3 As shown in the figure, the mine stress prediction process is divided into three stages: First, in the detection preparation stage, the determination of shot points, sensor deployment, anchor stress gauge deployment, connection and debugging of the entire system, and geological information recording are carried out; then, in the information acquisition stage, multi-pulse source excitation, seismic wave data collection, and strain data acquisition are performed; finally, in the data processing and prediction results stage, imaging structural surface and wave velocity inversion, stress data processing, multi-source data fusion, deep learning modeling and prediction, and construction of stress distribution field are carried out in sequence.

[0074] Specifically, such as Figure 4 As shown, the multi-pulse seismic mine stress prediction method includes the following steps S100-S400.

[0075] S100: Detection preparation.

[0076] like Figure 5 As shown, detection preparation can be achieved through the following steps S101-S104.

[0077] S101: Based on the spatial structure and geological conditions of the mine roadways, determine the deployment method of multipulse seismic sources and select appropriate excitation parameters such as excitation intensity and multipulse data.

[0078] In some embodiments, the method for determining the layout of multi-pulse seismic sources includes the layout of blast points: the distance between the blast hole and the loosened zone of the surrounding rock should be greater than 0.5m. The drilling direction should be perpendicular to the tunnel wall, the hole depth should be controlled between 2.0 and 2.5m, and the angle deviation should not exceed ±5°. The height of the hole position from the floor should be 1.2 to 1.5m, avoiding the support anchor bolts, with a avoidance distance of not less than 0.5m, to reduce the interference of the anchor bolt stress field. The charge amount of the electronically coded blasting source should be strictly limited to 50±2g, in accordance with the requirements of the blasting safety regulations. The hole should be sealed with compacted yellow mud, and the sealing density should not be less than 1.8g / cm³. 3 To ensure good energy coupling, seismic sources should be deployed on one side only, with a standard spacing of 20-30 meters, adjusted according to the actual detection resolution. In areas with faults, fracture zones, or other adverse geological conditions, the deployment density should be increased to a spacing of 15 meters.

[0079] S102: The measuring anchors are distributed on both sides of the roadway according to the arch shoulder and arch bottom in a certain form. The stress gauge on the measuring anchor is placed at the front end, and the high-sensitivity detector is placed at the end of the measuring anchor and exposed. During installation, it is necessary to ensure that the anchor is in close contact with the surrounding rock and the detector is in close contact with the anchor to improve the signal reception quality.

[0080] Step S102 above mainly realizes the placement of electronic components (including detectors and stress gauges). In some embodiments, the electronic components can be placed in the following manner:

[0081] Three-component geophones are arranged along one side of the roadway, with a frequency response range of 0.1 to 500 Hz. The spacing between the geophones is uniformly fixed at 10 ± 0.5 m to improve positioning accuracy and data consistency. Within a structural surface with a length of 200 m, the number of geophones should be no less than 10 to ensure sufficient sampling and response identification of key geological structures. (3) Installation of anchor stress gauges: In the key area of ​​the roadway, a set of support anchors is selected every 10 meters. Each set should have no less than five anchor stress gauges. Two high-precision strain gauges (range covering -105 to +105 MPa, accuracy ± 0.1 MPa) are symmetrically pasted on the free section of the anchor (50 cm away from the anchor tray). At the same time, temperature compensation plates are added to eliminate the influence of thermal temperature changes. When installing multi-pulse seismic sensors, the ends of the anchor bolts must be ground smooth. At least 60% of the anchor bolts should be driven deeper than 3 meters into hard rock layers. Avoid loose or fractured areas in the tunnel. Three-component sensors are connected to the ends of the anchor bolts via a rigid transition band, ensuring consistent orientation: the x-axis points towards the excavation direction, the y-axis is horizontal and perpendicular to the rock wall, and the z-axis is vertically downwards. Simultaneously, synchronous triggering of all equipment is achieved via network protocol / fiber optic dual-mode time synchronization (at the moment of seismic source excitation, the geophone and strain gauge synchronously start acquiring data within 0.1 μs).

[0082] S103: Connect the detector and stress gauge end to the acquisition system to perform full system connection and debugging to ensure smooth signal transmission and equipment stability.

[0083] It should be noted that the connection method is described in Example 1, wherein the detector can be selected as a three-component detector, and the acquisition system includes an acquisition station and a field host.

[0084] S104: Record basic data such as spatial geometry of mine roadways, surrounding rock type, structural features, anchor bolt arrangement and depth, and record them in the geological database as the geological information basis for subsequent inversion and imaging structural surfaces.

[0085] S200: Signal Acquisition.

[0086] like Figure 6 As shown, signal acquisition can be achieved through the following steps S201-S202.

[0087] S201: According to the predetermined excitation scheme, a multi-pulse seismic source slot wave signal is generated, and the detector simultaneously receives and records the reflected seismic wave signal. Multiple excitation rounds are performed within each acquisition cycle to improve the signal-to-noise ratio and store complete seismic waveform data to ensure stable and reliable signal quality.

[0088] Step S201 is used to collect seismic wave data. In some embodiments, seismic wave data can be collected by using a three-component detector to receive three-component waveforms, focusing on receiving the X, Y, and Z components, while recording the arrival time, polarization, amplitude, frequency, and anchor position of the direct and reflected waves.

[0089] S202: The strain gauge at the top of the anchor bolt records the anchor bolt strain signal in real time, and records its minute change trend through high-frequency sampling; the acquisition system processes and stores the strain data synchronously to ensure data integrity and continuity, and obtains high-precision time information through network protocols (such as NTP), and records seismic data and strain information synchronously to provide a reliable observation data foundation for geostress inversion.

[0090] S300: Data processing.

[0091] like Figure 7 As shown, signal acquisition can be achieved through the following steps S301-S302.

[0092] S301: Seismic data imaging structural surfaces and wave velocity inversion.

[0093] By analyzing the kinematic characteristics of reflected waves, such as propagation time, waveform amplitude, and frequency, the reverse time migration method is used to locate the internal structural surfaces, adverse geological bodies, and their geometric shapes within the tunnel. A tomographic inversion algorithm is then used to construct a velocity distribution map of the underground structural surfaces, providing velocity and structural surface parameters for geostress modeling.

[0094] In some embodiments, the specific methods for inverting seismic data imaging structural surfaces and wave velocities are as follows:

[0095] Noise filtering and signal enhancement are performed on the three-component geophone data. The initial velocity profile is obtained by inverting the velocity field using travel-time tomography. The surrounding rock velocity parameters are then finely inverted using the full waveform inversion method. Finally, the structural surface is imaged using reverse time migration. A four-step method is proposed to invert wave velocity and image the structural surface. First, the anisotropy parameters of the reflected wave dispersion curve are analyzed to extract the reflected wave components. Then, the initial velocity field is constructed using the first arrival wave travel-time tomography method. The velocity field is then corrected with high precision using full waveform inversion. The formula is as follows (1). Finally, the structural surface is finely imaged using reverse time migration.

[0096]

[0097] In the formula:

[0098] Sobs(t): Full waveform (including direct wave, reflected wave, and scattered wave);

[0099] Forward modeling operator (finite difference decomposition of acoustic / elastic wave equations);

[0100] β: Regularization weight;

[0101] T: Total duration of seismic wave acquisition;

[0102] v0: Initial wave velocity model (travel-time tomography inversion wave velocity);

[0103] v: v = v0 + Δv (interface wave velocity changes gradient);

[0104] v(x): Final wave velocity field.

[0105] In some embodiments, the method for realizing seismic data imaging of structural surfaces and wave velocity inversion includes: actively probing the surrounding rock structure of the mine using multi-pulse excitation signals, collecting reflected wave data using a geophone, establishing a multi-parameter fitting calculation for tomography based on information such as propagation time, amplitude, and frequency, and reconstructing the initial velocity information of the subsurface medium. Based on this, a full-waveform inversion algorithm is used to obtain a relatively accurate distribution of wave velocities at geological interfaces at different depths. Finally, reverse time migration is used to achieve imaging of the structural surfaces.

[0106] S302: Ground stress data processing.

[0107] The strain gauge data is preprocessed (median filtering) to eliminate outliers. Based on the elastic parameters and geometric characteristics of the anchor bolt, the local stress values ​​of the surrounding rock mass are inferred using the stress-strain relationship. The measured stress data from multiple anchor bolt measuring points are then aggregated to provide spatial calibration and boundary constraints for the deep learning model.

[0108] In some embodiments, the process of stress back-calculation (i.e., applying the stress-strain relationship to back-calculate the local stress value of the rock mass surrounding the anchor bolt) is as follows:

[0109] Based on the strain data of the strain gauge recorded in real time by the anchor stress gauge, the stress is calculated using formula (2), which provides point stress data for subsequent full-space geostress prediction.

[0110] σ=E*ε (2)

[0111] In the formula, σ is stress, E is elastic modulus, and ε is strain.

[0112] In some embodiments, the method for processing geostress data includes: installing multiple sets of strain gauges on the mine support anchor bolts and collecting strain data at different time points and under different stress states. Combining the mechanical parameters of the anchor bolts (such as Young's modulus, cross-sectional area, etc.), the stress state of the anchor bolts is inversely deduced based on the stress-strain formula. The obtained local stress values ​​constitute a spatial scatter set, providing high-precision input for subsequent geostress field prediction and serving as a constraint term for the deep learning model.

[0113] S400: Ground stress prediction.

[0114] In this embodiment, geostress prediction can be achieved as follows: the inverted surrounding rock velocity results, structural surface distribution information, and measured stress values ​​of anchor bolts are used as input vectors. A semi-supervised learning deep neural network model is used to construct a mapping relationship based on the geostress at a single point on the anchor bolt, the wave velocity of the medium, and the imaging structural surface. The geostress at a single point on the anchor bolt and the wave velocity at that location are used as constraints, and the imaging direction of the structural surface is incorporated as a reference to establish a quantitative prediction of the magnitude of geostress in the entire underground space.

[0115] In some embodiments, geostress prediction can be achieved by constructing a deep learning geostress prediction model, such as a convolutional neural network (CNN) or a graph neural network (GNN). Using the inverted seismic wave velocity distribution and anchor stress data from the imaging structural surface as input features, a semi-supervised deep learning network is constructed. The magnitude of the geostress at the anchor and the imaging structural surface are used as the supervision basis, while geostress prediction at other locations is based on unsupervised learning of the imaging structural surface and wave velocity, thus achieving semi-supervised geostress prediction. A three-dimensional spatial interpolation method is used to spatially connect the point-like stress data, constructing a geostress distribution field. Ultimately, the coupling relationship between geological structure and wave velocity and geostress learned by the model is realized, quantitatively predicting geostress throughout the entire space of the mine, improving the prediction accuracy and full spatial coverage of the stress field.

[0116] like Figure 8 The figure shown is a schematic diagram of the geostress prediction results obtained based on the above steps S100-S400. Figure 8 The 3D model uses cubes to simulate the rock mass space surrounding the tunnel. Different filling textures within the cubes differentiate stress zones, representing the stress distribution around the tunnel (arc-shaped outline). Three types of stress zone textures are defined: dotted textures correspond to low-stress zones, representing relatively low-pressure areas near the tunnel after stress release; diagonal textures correspond to stress gradient zones, representing transitional zones between low and high stress, where stress changes are more pronounced; and mesh textures correspond to high-stress zones, representing original high-stress areas far from the tunnel and undisturbed by excavation. Figure 8 The entire structure is presented from a three-dimensional cross-sectional perspective, which visually reveals the spatial distribution characteristics of the low-stress zone, stress gradient zone, and high-stress zone of the surrounding rock after tunnel excavation.

[0117] Example 3:

[0118] This application provides a multi-pulse seismic mine stress prediction device, such as... Figure 9 As shown, the device includes a multi-pulse seismic detection structure 901, a signal acquisition module 902, a data processing module 903, and a ground stress prediction module 904 as described in any embodiment of Embodiment 1, wherein:

[0119] The signal acquisition module 902 is configured to generate multi-pulse seismic source slot wave signals according to determined excitation parameters. A three-component geophone receives and records the seismic wave signals. In each acquisition cycle, multiple rounds of excitation are performed and complete seismic waveform data is stored. Anchor bolt strain signals are recorded in real time by a stress gauge to obtain strain information. The on-site host synchronously processes and stores the strain data. The complete seismic waveform data and strain information are synchronously recorded through network protocol time synchronization.

[0120] The data processing module 903 is configured to locate the internal structural surfaces and adverse geological bodies of the tunnel based on the complete seismic waveform data using reverse time migration, construct a velocity distribution map of the underground structural surfaces using a tomographic inversion algorithm, perform median filtering preprocessing on the strain information, and infer the local stress value of the rock mass around the anchor bolt based on the elastic parameters and geometric characteristics of the anchor bolt through the stress-strain relationship, and summarize the measured stress data of multiple anchor bolt measuring points.

[0121] The ground stress prediction module 904 is configured to take the inverted seismic wave velocity distribution, imaging structural surface and anchor stress data as input features, and use a semi-supervised learning deep neural network model to construct the mapping relationship between anchor single-point ground stress, medium wave velocity and imaging structural surface. With anchor single-point ground stress and corresponding wave velocity as constraints, and incorporating the imaging direction of the structural surface, a quantitative prediction of underground full-space ground stress is established.

[0122] It should be noted that this multi-pulse seismic mine stress prediction device and the prior multi-pulse seismic mine stress prediction method share the same technical concept, have the same principle, and can achieve the same technical effect, so they will not be described in detail here. In some embodiments, the signal acquisition module 902 can be configured in the acquisition station 6, and the data processing module 903 and the stress prediction module 904 can be configured in the field host 1.

[0123] The above embodiments are only used to illustrate this application and are not intended to limit this application. Those skilled in the art can make various changes and modifications without departing from the spirit and scope of this application. Therefore, all equivalent technical solutions also fall within the scope of this application, and the patent protection scope of this application should be defined by the claims.

Claims

1. A method for predicting in-situ stress in mines based on multi-pulse seismic detection structures, characterized in that, The multi-pulse seismic detection structure includes a field host, a multi-pulse source, cables, a three-component geophone, a preamplifier, an acquisition station, an anchor tray, and anchors. The multi-pulse source is connected to the field host via cables, and its boreholes are sealed with yellow mud. The three-component geophone is rigidly connected to the end of the anchor through the anchor tray, and is sequentially connected to the preamplifier and the acquisition station, ultimately connecting to the field host. A stress gauge is installed on the free section of the anchor, and the stress gauge includes a ceramic substrate, a temperature compensation plate, a resistance strain gauge, and waterproof adhesive. The resistance strain gauge and the temperature compensation plate are fixed to the ceramic substrate with the waterproof adhesive and are connected to the signal adjustment module via cables. The signal adjustment module is connected to the acquisition station via a cable; The method includes: Detection Preparation: Based on the spatial structure and geological conditions of the mine roadway, determine the layout and excitation parameters of the multi-pulse seismic source; the excitation parameters include excitation intensity and multi-pulse data; distribute the anchor bolts along the shoulders and bottom of the roadway on both sides, place the stress gauge at the front end of the anchor bolt, and place the high-sensitivity geophone at the end of the anchor bolt and expose it, ensuring close contact between the anchor bolt and the surrounding rock, and between the geophone and the anchor bolt; connect the three-component geophone and the stress gauge end to the preamplifier and signal adjustment module respectively, and perform system connection and debugging; record the spatial geometry information of the mine roadway, the type of surrounding rock, structural features, anchor bolt layout and depth, and store them in the geological database; Signal acquisition: A multi-pulse seismic source channel wave signal is generated according to the determined excitation parameters. A three-component geophone receives and records the seismic wave signal. Multiple excitations are performed in each acquisition cycle and complete seismic waveform data is stored. The strain signal of the anchor bolt is recorded in real time by a stress gauge to obtain strain information. The on-site host synchronously processes and stores the strain data. The complete seismic waveform data and strain information are synchronously recorded through network protocol time synchronization. Data processing: Based on the complete seismic waveform data, the internal structural surfaces and adverse geological bodies of the tunnel are located using reverse time migration, and the velocity distribution map of the underground structural surfaces is constructed using tomographic inversion algorithm; the strain information is preprocessed by median filtering, and the local stress value of the rock mass around the anchor is inferred from the stress-strain relationship based on the elastic parameters and geometric characteristics of the anchor, and the measured stress data of multiple anchor measuring points are summarized. Ground stress prediction: Using the inverted seismic wave velocity distribution, imaging structural surface, and anchor stress data as input features, a semi-supervised learning deep neural network model is used to construct the mapping relationship between anchor single-point ground stress, medium wave velocity, and imaging structural surface. With anchor single-point ground stress and corresponding wave velocity as constraints, the imaging direction of the structural surface is incorporated to establish a quantitative prediction of ground stress in the entire underground space. Based on the complete seismic waveform data, reverse time migration is used to locate internal structural surfaces and adverse geological bodies within the tunnel. A velocity distribution map of the subsurface structural surfaces is constructed using a tomographic inversion algorithm, including: The seismic wave signal is subjected to noise filtering, and the anisotropy parameters of the dispersion curve are extracted; The components of the reflected wave are extracted based on the anisotropic parameters of the reflected wave dispersion curve, and the initial wave velocity model is constructed using the first arrival wave travel time tomography method. Based on the initial velocity field, the wave velocity distribution is corrected using formula (1), and fine imaging of the structural surface is achieved using reverse time migration technology: ； In the formula, : Full waveform, including direct wave, reflected wave and scattered wave; : Positive operator; Regularization weights; Total seismic wave acquisition time; Initial wave velocity model; : , The total velocity field to be inverted. For speed increments; : Final wave velocity field.

2. The multi-pulse seismic mine stress prediction method as described in claim 1, characterized in that, The distance between the blast hole of the multi-pulse seismic source and the loosened zone of the surrounding rock is greater than 0.5m. The drilling direction is perpendicular to the tunnel wall, the hole depth is controlled between 2.0 and 2.5m, the angle deviation does not exceed ±5°, the hole position is 1.2 to 1.5m above the floor, avoiding the support anchor bolts with a clearance distance of not less than 0.5m, the charge of the electronically coded blasting source is 50±2g, and the hole is sealed with compacted yellow mud with a sealing density of not less than 1.8g / cm³. 3 The seismic source is deployed on one side, with a standard spacing of 20-30m, and the spacing is increased to 15m in fault and fracture zone areas.

3. The multi-pulse seismic mine stress prediction method as described in claim 1, characterized in that, The three-component detector is connected to the anchor plate via a rigid transition band, and its orientation is fixed as follows: x The axis points in the direction of excavation. y The axis is horizontal and perpendicular to the rock wall. z The axis is vertically downward.

4. The multi-pulse seismic mine stress prediction method as described in claim 1, characterized in that, The acquisition station achieves synchronous triggering with the multipulse source, three-component detector, and signal adjustment module through fiber optic / network protocol dual-mode time synchronization, with a delay of ≤0.1μs, and synchronously stores seismic waveform data and anchor strain data.

5. The multi-pulse seismic mine stress prediction method as described in claim 1, characterized in that, The ends of the anchor bolts are ground smooth and penetrate ≥3m into hard rock layers, accounting for ≥60%.

6. The multi-pulse seismic mine stress prediction method as described in claim 1, characterized in that, It also includes a controller, which, based on the anchor bolt's elastic parameters and geometric characteristics, uses the stress-strain relationship to infer the local stress value of the rock mass surrounding the anchor bolt. The calculation formula is as follows: σ=E*ε (2) In the formula, σ is stress, E is elastic modulus, and ε is strain.

7. The multi-pulse seismic mine stress prediction method as described in claim 1, characterized in that, Using the inverted seismic wave velocity distribution, imaging structural surface, and anchor stress data as input features, a semi-supervised deep neural network model is employed to construct a mapping relationship between anchor stress at a single point, medium wave velocity, and the imaging structural surface. With anchor stress at a single point and corresponding wave velocity as constraints, and incorporating the imaging direction of the structural surface, a quantitative prediction of subsurface stress across the entire space is established, including: Constructing a semi-supervised learning deep neural network model; The inverted seismic wave velocity distribution, imaging structure surface, and anchor stress data are used as input features. The magnitude of the ground stress at the monitored anchor and the imaging structure surface are used as the supervision basis, while the ground stress at the non-monitored anchor is learned based on the imaging structure surface and wave velocity, thus realizing semi-supervised ground stress prediction. A three-dimensional spatial interpolation method is used to make the predicted point stress data continuous in space and construct the geostress distribution field.

8. A multi-pulse seismic mine stress prediction device, based on the method as described in any one of claims 1 to 7, characterized in that, It includes a multi-pulse seismic detection structure, a signal acquisition module, a data processing module, and a geostress prediction module, among which: The signal acquisition module is configured to generate multi-pulse seismic source slot wave signals according to determined excitation parameters. A three-component geophone receives and records the seismic wave signals. In each acquisition cycle, multiple rounds of excitation are performed and complete seismic waveform data is stored. Anchor bolt strain signals are recorded in real time by a stress gauge to obtain strain information. The on-site host synchronously processes and stores the strain data. The complete seismic waveform data and strain information are synchronously recorded through network protocol time synchronization. The data processing module is configured to locate the internal structural surfaces and adverse geological bodies of the tunnel using reverse time migration based on the complete seismic waveform data, construct the velocity distribution map of the underground structural surfaces using tomographic inversion algorithm, perform median filtering preprocessing on the strain information, and infer the local stress value of the rock mass around the anchor bolt based on the elastic parameters and geometric characteristics of the anchor bolt through the stress-strain relationship, and summarize the measured stress data of multiple anchor bolt measuring points. The geostress prediction module is configured to take the inverted seismic wave velocity distribution, imaging structural surface, and anchor stress data as input features, and use a semi-supervised learning deep neural network model to construct the mapping relationship between anchor single-point geostress, medium wave velocity, and imaging structural surface. With anchor single-point geostress and corresponding wave velocity as constraints, and incorporating the imaging direction of the structural surface, a quantitative prediction of geostress in the entire underground space is established.

Images